2D Potential Problems Research Articles

The Kriging integration method applied to the boundary element analysis of Poisson problems

A novel efficient technique is presented for the evaluation of domain integrals that appear in the boundary element method (BEM). Herein, the source term is approximated with the use of radial basis functions, as in the dual reciprocity BEM. The proposed technique, called the Kriging Integration Method (KIM), comprises the use of the Simple Kriging Method in non-overlapping patches for obtaining the weights of the integration points located inside. As it is necessary to compute the integrals of the covariance function prior to obtaining these weights, this can be efficiently realized using the Cartesian Transformation Method. The domain integrals over all the generated partitions are then computed and added to obtain the value of the whole-domain integral. Using KIM, it is possible to evaluate approximately weakly singular domain integrals over simply or multiply connected domains without applying any transformation or regularization method to the singular integrand. The numerical results obtained in several 2D potential problems demonstrate that this integration scheme is as accurate as both the dual reciprocity method and RIM and less time consuming than the RIM.

The semianalytical analysis of nearly singular integrals in 2D potential problem by isogeometric boundary element method

AbstractBenefited from the accuracy improvement in modeling physical problem of complex geometry and integrating the discretization and simulation, the isogeometric analysis in boundary element method (IGABEM) has been drawn a great deal of attention. The nearly singular integrals of 2D potential problem in the IGABEM are addressed by a semianalytical scheme in the present work. We use the subtraction technique to separate the integrals to singular and nonsingular parts, where the singular parts can be calculated by the analytical formulae derived by utilizing a series of integration by parts, while the nonsingular parts are calculated numerically with fewer quadrature points. Comparing the present semianalytical results with the ones of exact solutions, we find that the present method can obtain precise potential and flux densities of inner points much closer to the boundary without refining the elements nearby. Sufficient comparisons with other regularization schemes, such as the exponential and sinh transformation methods, are also conducted. The results in the numerical examples show the competitiveness of the present method, especially when calculating the nearly strongly and highly singular integrals during the simulation of the flux density.

The Improved Interpolating Complex Variable Element Free Galerkin Method for Two-Dimensional Potential Problems

In this paper, the improved interpolating complex variable moving least squares (IICVMLS) method is applied, in which the complete basis function is introduced and combined with the singular weight function to achieve the orthometric basis function. Then, the interpolating shape function is achieved to construct the interpolating trial function. Incorporating the IICVMLS method and the Galerkin integral weak form, an improved interpolating complex variable element free Galerkin (IICVEFG) method is proposed to solve the 2D potential problem. Because the essential boundary conditions can be straightaway imposed in the above method, the expressions of final dispersed matrices are more concise in contrast to the non-interpolating complex variable meshless methods. Through analyzing four specific potential problems, the IICVEFG method is validated with greater computing precision and efficiency.

A Complex Variable Boundary Element-Free Method for Potential and Helmholtz Problems in Three Dimensions

The complex variable boundary element-free method (CVBEFM) is a meshless method that takes the advantages of both boundary integral equations (BIEs) in dimension reduction and the complex variable moving least squares (CVMLS) approximation in element elimination. The CVBEFM is developed in this paper for solving 3D problems. This paper is an attempt in applying complex variable meshless methods to 3D problems. Formulations of the CVMLS approximation on 3D surfaces and the CVBEFM for 3D potential and Helmholtz problems are given. In the current implementation, the CVMLS shape function of 3D problems is formed with 1D basis functions, and the boundary conditions in the CVBEFM can be applied directly and easily. Some numerical examples are presented to demonstrate the method.

The dimension splitting reproducing kernel particle method for three‐dimensional potential problems

SummaryIn this paper, the dimension splitting reproducing kernel particle method (DSRKPM) for three‐dimensional (3D) potential problems is presented. In the DSRKPM, a 3D potential problem can be transformed into a series of two‐dimensional (2D) ones in the dimension splitting direction. The reproducing kernel particle method (RKPM) is used to solve each 2D problem, the essential boundary conditions are imposed by penalty method, and the discretized equation is obtained from Galerkin weak form of potential problems. Finite difference method is used in the dimension splitting direction. Then, by combining a series of the equations of the RKPM for solving 2D problems, the final equation of the DSRKPM for 3D potential problems is obtained. Five example problems on regular or irregular domains are selected to show that the DSRKPM has higher computational efficiency than the RKPM and the improved element‐free Galerkin method for 3D potential problems.

Fast direct isogeometric boundary element method for 3D potential problems based on HODLR matrix

A novel fast direct solver based on isogeometric boundary element method (IGABEM) is presented for solving 3D potential problems, which uses the hierarchical off-diagonal low-rank (HODLR) matrix structure arising from the discretization of boundary integral equations. Since the HODLR matrix can be factored into the product form of some diagonal blocks, we can use the Sherman–Morrison–Woodbury formula to solve the inverse of a HODLR matrix efficiently. For large scale problems, an accelerated adaptive cross approximation algorithm is developed to decompose the off-diagonal submatrices. In numerical implementation, bivariate NURBS basis functions are used to describe the geometry. Meanwhile, the same NURBS basis functions are also used to approximate the unknown boundary quantities. The present method is applied to some numerical examples, including an infinite space containing twenty spherical cavities. The numerical results clearly show that the fast direct solver developed in the paper can obtain accurate results with less CPU time.

A multi-level technique for the Method of Fundamental Solutions without regularization and desingularization

The traditional Method of Fundamental Solutions is revisited, based on a special strategy of defining the external source points. Unlike the classical Method of Fundamental Solutions, the sources are categorized into groups; the density of the spatial distribution of the sources decreases rapidly far from the boundary. On each group, the original problem is discretized using the same set of boundary collocation points. Such groups of sources are constructed in a fully automated way by the quadtree/octtree algorithm. The discretized problems are solved in the sense of least squares. A simple multi-level method is built up, using the (conjugate) gradient iteration as a smoothing procedure. The resulting method significantly reduces the computational complexity. Moreover, the problem of evaluation singular integrals as well as the problem of severely ill-conditioned matrices are avoided. The method is generalized to 3D axisymmetric potential problems as well.

An improved local boundary integral equation method implemented by the transformed MLS approximation with the delta property

The meshless local boundary integral equation (LBIE) method is a promising method for solving problems of elasticity with nonhomogeneous material properties. However, the shape functions of the LBIE method obtained by the moving least-squares (MLS) approximation, generally, do not satisfy the Kronecker delta property. To impose boundary conditions, numerical integrations of LBIE method are carried out under consideration of boundary conditions and the fictitious nodal values would be solved out. In this paper, the transformed MLS shape functions can be obtained, satisfying the Kronecker delta property and requiring no singular weight function. And then the potential and flux in 2D potential problems would be regarded as independent variables each other. The numerical integrations of LBIEs can be processed without regard to boundary conditions and the coefficient matrix of LBIE method would be irrelated to boundary conditions. Finally boundary conditions would be directly imposed and the unknown nodal potentials or the unknown fluxes on the boundary nodes would be solved out. Three numerical examples are computed to verify this feasibility. The coincidence of the numerical results obtained by the proposed method with the traditional LBIE method shows the feasibility of the proposed method.

A NURBS-enhanced improved interpolating boundary element-free method for 2D potential problems and accelerated by fast multipole method

Isogeometric analysis (IGA) has been widely applied in the finite element method and boundary element method (BEM), in which the geometry discretization error can be avoided. This paper proposes a NURBS-enhanced meshless improved interpolating boundary element-free method for 2D potential problems. In the proposed method, non-uniform rational B-spline (NURBS) basis functions are applied to reproduce the geometry like in IGA, and the boundary integral cells, called isogeometric cells, are defined by the knot vector of NURBS in parameter space. Thus, the geometry can remain the same at all stages because refining a NURBS curve will not change its shape. An improved interpolating moving least-square (IIMLS) method is applied to approximate the field in parameter space, and the boundary nodes can be defined using Greville abscissae definition. Compared with IGA in BEM, the shape functions obtained by IIMLS in the proposed method have the delta function property, and the boundary conditions can be applied directly. In addition, most methods for the treatment of the singular integrals in BEM can be applied easily in the proposed method. Fast multipole method is further coupled with the proposed method for large-scale computation.

Evaluating the Origin Intensity Factor in the Singular Boundary Method for Three-Dimensional Dirichlet Problems

AbstractIn this paper, a new formulation is proposed to evaluate the origin intensity factors (OIFs) in the singular boundary method (SBM) for solving 3D potential problems with Dirichlet boundary condition. The SBM is a strong-form boundary discretization collocation technique and is mathematically simple, easy-to-program, and free of mesh. The crucial step in the implementation of the SBM is to determine the OIFs which isolate the singularities of the fundamental solutions. Traditionally, the inverse interpolation technique (IIT) is adopted to calculate the OIFs on Dirichlet boundary, which is time consuming for large-scale simulation. In recent years, the new methodology has been developed to efficiently calculate the OIFs on Neumann boundary, but the Dirichlet problem remains an open issue. This study employs the subtracting and adding-back technique based on the integration of the fundamental solution over the whole boundary to develop a new formulation of the OIFs on 3D Dirichlet boundary. Several problems with varied domain shapes and boundary conditions are carried out to validate the effectiveness and feasibility of the proposed scheme in comparison with the SBM based on inverse interpolation technique, the method of fundamental solutions, and the boundary element method.

Improved complex variable moving least squares approximation for three-dimensional problems using boundary integral equations

The improved complex variable moving least squares approximation is an efficient method to generate meshless approximation functions. In the past, the approximation has been used only for 2D problems. In this paper, the approximation is developed to solve 3D problems. Theoretical error estimation of the approximation is given. Then, incorporating the approximation into boundary integral equations, a symmetric and boundary-only meshless method, the complex variable Galerkin boundary node method, is developed and analyzed theoretically for 3D potential, Helmholtz and Stokes problems. Numerical results demonstrate the accuracy and efficiency of the developed method.

NURBS-enhanced boundary element method based on independent geometry and field approximation for 2D potential problems

Non-uniform rational B-spline (NURBS) in Isogeometric analysis (IGA) is coupled with the boundary element method (BEM) for 2D potential problems in this paper. The geometry and field are usually approximated by the same basis functions in IGA, such as the B-spline or the NURBS basis functions. In the proposed method, these two kinds of approximation are performed independently, i.e. the geometry is reproduced by the NURBS basis functions while the field is approximated by the traditional Lagrangian basis functions which are used in the conventional BEM. The proposed method has the advantage that the geometry can be reproduced exactly at all stages in IGA methods. Actually, one can use the computer aided design (CAD) software or NURBS library to perform the operations related to the geometry. The field approximation is performed in parameter space and separated from the geometry. Thus, it can be implemented easily as the conventional BEM since most algorithms for BEM can be applied directly, such as the methods for treatment of the singular integrals, and the boundary conditions can be imposed directly. Numerical examples have demonstrated the accuracy of the proposed method.

A supernodal block factorized sparse approximate inverse for non-symmetric linear systems

The concept of supernodes, originally developed to accelerate direct solution methods for linear systems, is generalized to block factorized sparse approximate inverse (Block FSAI) preconditioning of non-symmetric linear systems. It is shown that aggregating the unknowns in clusters that are processed together is particularly useful both to reduce the cost for the preconditioner setup and accelerate the convergence of the iterative solver. A set of numerical experiments performed on matrices arising from the meshfree discretization of 2D and 3D potential problems, where a very large number of nodal contacts is usually found, shows that the supernodal Block FSAI preconditioner outperforms the native algorithm and exhibits a much more stable behavior with respect to the variation of the user-specified parameters.

A fast multipole implementation of the simplified hybrid boundary element method: application to 2D potential problems

The ultimate subject of this work is the implementation and testing of a novel numerical tool that can simulate on a personal computer and only in a few minutes a problem with many millions of degrees of freedom. The authors have already successfully developed and tested a technique that turned out to be a modified, reverse fast-multipole implementation for the conventional BEM. The variationally based hybrid BEM leads to a computationally less intensive formulation than in the conventional BEM for large-scale 2D and 3D problems of potential and elasticity. This formulation is especially advantageous for problems of complicated geometry and topology or requiring complicated fundamental solutions. The proposed implementation of the fast multipole method (FMM) for the simplified, hybrid BEM deals with the transpose of the double-layer potential matrix as well as with the nodal matrix expression of the potential fundamental solution. The basic aspects of the FMM are firstly introduced for the conventional BEM as well as for its expedite version. This takes most part of the present paper, which ends up with some validating numerical results. The FMM outline for the simplified hybrid BEM is shown in a separate section, as its numerical implementation is still in progress.

An adaptive cell-based domain integration method for treatment of domain integrals in 3D boundary element method for potential and elasticity problems

An adaptive cell-based domain integration method (CDIM) is proposed for the treatment of domain integrals in 3D boundary element method (BEM). The domain integrals are computed in background cells rather than volume elements. The cells are created from the boundary elements based on an adaptive oct-tree structure and no other discretization is needed. Cells containing the boundary elements are subdivided into smaller sub-cells adaptively according to the sizes and levels of the boundary elements; and the sub-cells outside the domain are deleted to obtain the desired accuracy. The method is applied in the 3D potential and elasticity problems in this paper.

An isogeometric boundary element method using adaptive integral method for 3D potential problems

Due to the merits of exact geometrical representation, high accuracy, no meshing process, and only the boundary description of the problem etc., the isogeometric boundary element method, i.e. IGABEM, has achieved rapid development. In this paper, based on the upper bound of the relative error of the Gaussian quadrature formula, we presented a 3D IGABEM using adaptive integration method for potential problems. This method offers a number of key improvements compared with conventional IGABEM. Firstly, the boundary element integrations can be computed easily and effectively at optimal computational cost. Secondly, the adaptive algorithm can cope with the common situation where the sizes of adjacent cells are significantly different. Moreover, the presented method can compute the nearly singular integrals easily, owing to the use of subdivision technique. To accurately evaluate the singular integrals appearing in our method, the power series expansion method is employed. The integration surface is on the real surface of the model, rather than the interpolation surface, i.e. no geometrical errors. Thus, the value of integral is more accurate than the traditional boundary element method, which can improve the computation accuracy of the IGABEM. Numerical tests show that the presented method has good performance in both exactness and convergence.

用于解析函数复分析的共轭边界元法

An analytic function is composed of 2 real conjugate harmonic functions, of which the complex analysis plays an important role in the fields of applied mathematics and mechanics. A set of weighted residual equations were proposed and proved to be equivalent to the approximate solution to the original problem involving 2 governing equations in the domain, the boundary condition and the CauchyRiemann equation at the boundary. 2 conventional direct boundary integral equations at the boundary collocation points were deduced from 2 of the weighted residual equations, and 1 finite difference equation was deduced from the rest one. The mathematical problem arising from the illconditioned linear equations was solved and the Cauchy integral equation was adopted for numerical calculation of the fields at the internal points inside the domain. Finally, the proposed conjugate boundary element method with constant elements was completely established. 3 examples demonstrate that, the proposed method is valid for analytic functions in terms of the power function, the exponential function and the logarithmic function in interior or exterior domains, and the error estimation of the proposed method is at the same order as that of the boundary element method for 2D potential problems.

A General 2D Meshless Interpolating Boundary Node Method Based on the Parameter Space

The presented study proposed an improved interpolating boundary node method (IIBNM) for 2D potential problems. The improved interpolating moving least-square (IIMLS) method was applied to construct the shape functions, of which the delta function properties and boundary conditions were directly implemented. In addition, any weight function used in the moving least-square (MLS) method was also applicable in the IIMLS method. Boundary cells were required in the computation of the boundary integrals, and additional discretization error was not avoided if traditional cells were used to approximate the geometry. The present study applied the parametric cells created in the parameter space to preserve the exact geometry, and the geometry was maintained due to the number of cells. Only the number of nodes on the boundary was required as additional information for boundary node construction. Most importantly, the IIMLS method can be applied in the parameter space to construct shape functions without the requirement of additional computations for the curve length.

Line integration method for treatment of domain integrals in 3D boundary element method for potential and elasticity problems

A line integration method is presented in this paper for evaluation of domain integrals in 3D problems. The method is a boundary-only discretization method and the domain integrals can be computed by sum of integrals on one-dimensional straight lines. Divergence theorem is used to transform the domain integrals into boundary integrals with one-dimensional integrals. The boundary integrals can be evaluated by boundary elements with integral points. Each integral point can be used to construct an integral line, and the domain integrals can be finally computed by line integrals on integral lines. Only the boundary discretization is needed and background cells are used to cut the integral lines into sub-lines to obtain the desired accuracy. The method is proved and applied in boundary element method for 3D potential and elasticity problems. Numerical examples have demonstrated the accuracy of the proposed method.

An isogeometric boundary element method for three dimensional potential problems

Isogeometric analysis (IGA) coupled with boundary element method, i.e. IGABEM, received a lot of attention in recent years. In this paper, we extend the IGABEM to solve 3D potential problems. This method offers a number of key improvements compared with conventional piecewise polynomial formulations. Firstly, the models for analysis in the IGABEM are exact geometrical representation no matter how coarse the discretization of the studied bodies is, thus the IGABEM ensures that no geometrical errors are produced in the analysis process. Secondly, a meshing process is no longer required, which means redundant computations are eliminated to allow analysis to be carried out with greatly reduced pre-processing. To accurately evaluate the singular integrals appearing in our method, the power series expansion method is employed. The integration surface is on the real surface of the model, rather than the interpolation surface, i.e. no geometrical errors. Thus, the value of integral is more accurate than the traditional boundary element method, which can improve the computation accuracy of the IGABEM. Some numerical examples for 3D potential problems are used to validate the solutions of the present method with analytical and numerical solutions available.